Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 2 3 [4] 5 6

Author Topic: Drunken masters - how to use the Teacher skill  (Read 29616 times)

Mudcrab

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #45 on: April 27, 2012, 01:09:00 pm »

Im trying this with three dwarfs right now and its right to say that demonstrations of whatever skill the dwarf with proficient teacher is proficient in lead to very fast skill gains. However they also often do other demonstrations, which seem to be way less effective on account of their low skill in the actual skill being trained.

Its a big downside to the plan :(
« Last Edit: April 27, 2012, 01:15:08 pm by Mudcrab »
Logged

Calech

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #46 on: April 27, 2012, 04:34:04 pm »

If one were to perform definitive science on this, what methods should be tested?

It seems there are three main variables: Applicable Skill, Teacher, and Student.  If we follow rationally, this allows for a "is skilled" or "no skill" on all three.  This means you need 4 teachers, with and without a teachable skill, and with and without the teaching skill, as well as two students, with and without the student skill.

This should occupy 16 dwarves total.
  • Skilled Axedwarf, Skilled Teacher, Skilled Student
  • Skilled Axedwarf, No Teacher, Skilled Student
  • Skilled Axedwarf, Skilled Teacher, No Student
  • Skilled Axedwarf, No Teacher, No Student
  • No Axedwarf, Skilled Teacher, Skilled Student
  • No Axedwarf, No Teacher, Skilled Student
  • No Axedwarf, Skilled Teacher, No Student
  • No Axedwarf, No Teacher, No Student
That is, (Axe+Teacher) and (Student) in two separate dwarves.

Proposed method is to embark with 6 dwarves and one laborer.  The 6 will take some combination of the above skills, allowing 3 instances to be tested at one embark.  Very quickly wall off, set barracks, and constant training for 1 year (perhaps from Summer to Summer to ensure that it's a solid year each) and compare the findings.  Perform this for each of the 8 instances and compare data.  Assuming the fortress isn't suddenly interrupted for some reason, this should allow for some fairly reliable results and solid evidence as to what matters where.

This has actually gotten me interested enough to try this out. I've been quite fond of bringing military instructors at embark, and it would be nice to know how best to choose them!

I've thrown together a test embark with (hopefully) enough food and drink to last seven dwarves for a year; any migrants will have to provide for themselves or be given a complementary slab inside the training area. The idea I'm using is to create a fixed starting point, ie. use the same Summer save to start with, and partition the dwarves into three squads along a different division each time, with possibly a control group of 6 single-dwarf squads (who should logically do Individual Combat Drills non-stop due to lack of other options).

The dwarves will be unmodded so their bodily needs may cause a little variance in the numbers, but since the same is true of a 'real' fortress anyway this simulates the practical case a little better.

It does appear that the biggest time-consumer in this is likely to be writing it all up; I've just done an overview of the situation at the end of the first Summer (ie. a quarter of the way through the experiment) and it's taken quite a while to extract the numbers from DT, copy them down, and type random waffling discussing the results. Still, hopefully the end results will prove interesting, unless my Concentration skill fails me by the end of it :)
Logged

Sadrice

  • Bay Watcher
  • Yertle et al
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #47 on: April 27, 2012, 10:34:04 pm »

Perhaps put them in two man squads, so that the student doesn't lead the demonstrations more than half the time?  Of course, the student would probably train the teacher up to dabbling in several useless skills, and then they would both give crappy lessons.
Logged

Calech

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #48 on: April 28, 2012, 03:12:21 pm »

Well, that took longer than expected.

I've uploaded the initial embark setup to DFFD (here) if anyone wants to set a scenario running using it, or just wants to see what happens if you wall out the immigrants and let them be savaged by honey badgers.

The first year of experiments has been run, and it doesn't really tell us anything particularly insightful or interesting, just that having more skill is better - which is sort of intuitively obvious, really.


Starting skill levels (start of summer year 1):
Spoiler (click to show/hide)

First test:

Axe/Teacher + Student
Axe + Student
Teacher + Unskilled

Summer, 1st year
Spoiler: Log (click to show/hide)

End of Summer report:
Spoiler: Skills (click to show/hide)

Conclusions, Summer, Year 1:

It looks like Concentration and Student train at the exact same rate, which is interesting - and the Teacher skill of the Teacher trains at the same rate as the Student/Concentration skills of the student. The Leader skill is included here as a minor interest, although a dwarf who isn't a captain has got 10xp of Leader somehow, even though none of the other 'members' did - not sure how he got this.

Squad 1 (with all the skills present) has quite clearly gained the most, with the Student in the squad gaining 6 levels of Axe skill in a single season - and the Teacher hasn't done too badly herself, gaining 3 levels in the same span of time.

Squad 3, with only a Teacher and no weapon skills to start, is suffering a little - they have attained the rank of Novice, but haven't gotten any further, and exhibit the lowest skill gains of the three squads (as you may expect).

Observer appears to be being trained here too; the Worker (who is not in the militia) has zero Observer skill. If Observer affects anything important, like spotting ambushers, it seems that fighting trains it - there doesn't seem to be an obvious link with the Student/Concentration skill gains so I'd guess it might not be linked to the demonstration aspect?


Autumn, 1st year
Spoiler: Log (click to show/hide)

End of Autumn report:
Spoiler: Skills (click to show/hide)

Conclusions, Autumn, Year 1:

Well, it looks like a few of the dwarves gave up on teaching - 3 of them gained no Teacher experience, and correspondingly their 3 opposite numbers gained no Student experience. The dwarves with higher skills gained even more, while the dwarves who are still on lower skill are gaining fewer points. Most of the dwarves are now showing as very agile or extremely agile, except Ral who is only 'slow to tire' (which isn't even agility based); I'm not entirely sure how much effect attributes have on the training process but if agility increase the rate of action, the higher agility dwarves will in theory pick up their skills faster, which should lead to a snowball effect in this case (train skill -> gain agility -> train skill faster -> gain more agility -> etc.) - it's just a question of how quickly you can get them to that 'critical mass'.


Winter, 1st year
Spoiler: Log (click to show/hide)

End of Winter report:
Spoiler: Skills (click to show/hide)

Conclusions, Winter, Year 1:

Happy New Year 251! Some interesting numbers this season. Squad 1 gained a huge amount of Teacher/Student experience (comparatively), but only improved by a lesser amount in their Axe/Fighting skills over the previous season, only gaining 4-8k experience compared to the 8-12k in the autumn. However their other fighting abilities, like Dodging and Wrestling, made some bigger gains, taking them to Novice in each.

Squad 2 is still not gaining any Teacher/Student experience - I wonder if they've actually been doing any demonstrations? - but have advanced by a decent amount nonetheless, gaining another ~4k experience in their major skills. Squad 3, which started with no weapon skills at all, is also only making progress at about the same rate as last season, but have been exercising that Teacher/Student relationship a bit - even so they have progressed at about the same rate as each other.

Spring, 2nd Year
Spoiler: Log (click to show/hide)

End of Spring report:
Spoiler: Skills (click to show/hide)

Conclusions, Spring, Year 2:

Mostly this is just the same story as before; the dwarves at higher skill levels are gaining skills faster, whilst the lower skilled
squad is struggling to catch up (but is gaining experience a bit faster now).

Also, Cog, the Student in Squad 2, displays a little anomaly - he has gained 264xp in Student, but 270xp in Concentration (and his captain gained 270xp in Teacher also). Wonder what happened here? If the training was cut short or interrupted somehow, only gains in Student were lost - the squad is not currently in a demonstration at the time of the save, either.

Dodging seems to have become the fashionable skill, as there were several observed Dodging demonstrations, and it seems that most of the dwarves are now Novice or better Dodgers - except Squad 3, who are still trailing along behind in all aspects.

Conclusions after one year training, first experiment:

The two squads of Axe/Student dwarves fared much better than the squad with only a Teacher - presumably, it's useful to have something to actually teach! Of the two squads to start with weapon skill, the Teacher/Student pair have become Legendary +5 in a year, whilst the Non-Teacher/Student pair are only just short of Legendary at the one-year mark.

All of the dwarves appear to have gained stats, although I don't have exact values and can't remember how they started out - Squad 1 have almost all the ultra-positive statements like 'basically unbreakable', Squad 2 are all high-positive, and Squad 3 are positive but not superlatively so.

Splitting the dwarves in a different pattern for the second experiment will give us more useful conclusions than simply "more is better". If the Axe/Teacher is paired with an Unskilled Dwarf, this should show how significant the Student skill is to the relationship, and coupling the Teacher and Student with no weapon skills should show if the combination can make up for the lack of weapon skill. The Axe-Only dwarf can then be partnered with the Worker (who is technically also Unskilled, as he has no weapon skills - Mining shouldn't be relevant if it's not an equipped weapon) to see what effect is gained from military-only skills with no academical capabilities. The other Student will therefore be left hauling the barrels around.
Logged

Calech

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #49 on: April 30, 2012, 10:56:26 am »

Apologies for double posting.

Results are in for the second and third experiments that I've run;

Experiment 2

Squad 1: Domas (Axe/Teacher) + Dastot (Unskilled)
Squad 2: Ral (Teacher) + iton (Student)
Squad 3: Nil (Axe) + Etur (Worker, unskilled)

Cog has Hauling duties this time.

To try and make the experiments work a little faster I did make one change, namely setting the FPS cap to 1000 instead of 100, so hopefully things will go a little quicker, although there is the potential for everything to go horribly wrong (but then what else is new?)

Spoiler: Summer log (click to show/hide)

End of Summer report:

Spoiler: Skills (click to show/hide)

Observations, Summer, 1st year:

Some curious results here. Squad 2 (Teacher + Student) are worse off than their counterparts from the first experiment (Squad 3 there, Teacher + Unskilled) in their primary skills - and show a rather unbalanced Teacher-Student partnership, with the Student gaining 6 times as much Teacher xp as the Teacher! Also it's interesting to see the non-round numbers again here, with Concentration at 'full' gain but the Teacher and Student both at 'partial'.

Squad 1 have also not performed as well here as their counterparts from Experiment 1, despite much higher gains in the academical skills - of course, in the first experiment, the equivalent squad already had Lvl 5 Teacher/Student skills, so (in theory!) they would have had better gains from the demonstrations they did do.

Squad 3 (Axe + Unskilled) have actually performed fairly comparably with their counterpart squad in Experiment 1 (Squad 2, Axe + Student). They had higher overall gains in the academical skills, and have finished at the same skill level in their main skills, albeit slightly behind in raw xp terms.

Spoiler: Rest of Year log (click to show/hide)

End of year report:

Spoiler: Skills (click to show/hide)

Conclusions, Second Experiment:

Just as before, Squad 1 contains our only Legendary Axedwarves. But while the pairing from the First Experiment had both reached Legendary +5 by now, here we've only reached Legendary +2 or +3, with the initial Student skill removed from the equation.

Squad 2 in this experiment (Teacher + Student) actually did pretty well for themselves. Comparing against the First Experiment, they did better than their counterpart squad (Teacher + Unskilled), and are only 1 or 2 levels behind the other squad (Axe + Student).

Squad 3 (Axe + Unskilled), comparing against the same two squads in the First Experiment, performed marginally better than (Teacher + Unskilled), being about 1 level higher, but worse than (Axe + Student), with a difference of about 4 levels there.

It appears, then, that having both academic skills present in a pairing is significant, as it seems to produce better skill gains over time than only having the Teacher skill (ref: Sqd 1/Exp 1 vs Sqd 1/Exp 2 and Sqd 3/Exp 1 vs Sqd 2/Exp 2), or not having the Teacher skill where the Student skill is present (ref: Sqd 1/Exp 1 vs Spd 2/Exp 1).

Surprisingly it's possible to train up from only the academical skills quite satisfyingly, as Sqd 2/Exp 2 attained better results than Sqd 3/Exp 2, despite Sqd 2/Exp 2 having no initial military experience.

From the original list of tests, this leaves only two untested combinations plus a couple of other interesting cases;

Unskilled + Student
Unskilled + Unskilled
Indvidual Combat Drills only (squad size 1)
Larger squad sizes (eg. Axe/Teacher + 2x Students)

Since I have a tragic lack of unskilled dwarves in the testing area, with only 2 of them, it makes some amount of sense to couple one of the remaining test cases with one of the other cases, to keep the dwarves occupied.


Experiment 3

Squad 1: Unskilled + Unskilled
Squad 2: 3-dwarf squad: Axe/Teacher, Student, Student 2

You kind of know the drill by now, migrants arrive, attacked by raging honey badgers, etc.

Spoiler: Log (click to show/hide)

End of year report:
Spoiler: Skills (click to show/hide)

Observations, Third Experiment:

So there's a couple of interesting points here. The Unskilled squad, who had no prior military or academic capability, attained Legendary +2, a result only bettered by two other combinations so far, namely Axe/Teacher + Student (Sqd 1/Exp 1) and Axe/Teacher + Unskilled (Sqd 1/Exp 2). They soundly outperformed even the Teacher + Student squad from Experiment 2 (Sqd 2/Exp 2).

Meanwhile, adding an extra dwarf to Squad 2 seemed to slow down individual skill gains somewhat, although two out of three are Legendary +1 after only a year's training with another dwarf at Master skill. The most stand-out number to me, though, is the Observer skill on the Axe/Teacher - your eyes do not decieve you, she gained 5 levels in Observer, the most I've seen in the course of the experiment. I suspect it may be linked to being attacked - in the 3-way sparring sessions, she was the target of 2 attackers, which may explain why her skill gains here are that much higher than others.

The total gain for Squad 2 in Teacher is less than the gain for Student/Concentration, so it may well be the case that the Teacher skill advances at the same rate irrespective of how many dwarves are being taught; note however that the intended Teacher gained far more Teacher skill than her subordinates (2350 > 650 + 760) so dedicated Teachers may be able to win out over dabblers.

The really unusual result here is that the completely unskilled squad has managed to do better than the squad with the military experience (Sqd 3/Exp 2), and beat out the squad with the academic qualifications (Sqd 2/Exp 2).
Logged

Loud Whispers

  • Bay Watcher
  • They said we have to aim higher, so we dug deeper.
    • View Profile
    • I APPLAUD YOU SIRRAH
Re: Drunken masters - how to use the Teacher skill
« Reply #50 on: April 30, 2012, 01:52:31 pm »

Conclusion: Teacher is one of the best Dorf misc skills in existence. And honey badger doesn't care.

Also that's not double posting, that's !!SCIENCE!!

jwest23

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #51 on: April 30, 2012, 02:23:20 pm »

The really unusual result here is that the completely unskilled squad has managed to do better than the squad with the military experience (Sqd 3/Exp 2), and beat out the squad with the academic qualifications (Sqd 2/Exp 2).

Thanks for taking the time to run the experiments and document the results.  Fantastic work!

My hunch is that the unusual result might have to do with the personalities of the dwarves involved.  I'm not in a position just now to load up the save and inspect it for myself.  Are the Sqd 3 / Exp 2 dwarves more motivated than the dwarves in Sqd 2 / Exp 2?

It might be worth re-running this particular scenario a few more times to eliminate any sort of odd personality bias.
Logged

Calech

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #52 on: April 30, 2012, 03:05:57 pm »

The really unusual result here is that the completely unskilled squad has managed to do better than the squad with the military experience (Sqd 3/Exp 2), and beat out the squad with the academic qualifications (Sqd 2/Exp 2).

Thanks for taking the time to run the experiments and document the results.  Fantastic work!

My hunch is that the unusual result might have to do with the personalities of the dwarves involved.  I'm not in a position just now to load up the save and inspect it for myself.  Are the Sqd 3 / Exp 2 dwarves more motivated than the dwarves in Sqd 2 / Exp 2?

It might be worth re-running this particular scenario a few more times to eliminate any sort of odd personality bias.

Good point; ideally for comparing the skill effects alone there would be 'standard' dwarves with identical stats and personalities, but I'm not sure that can even be modded / edited easily or if it would give meaningful results!

There's no specific 'motivation' personality token, but looking over the wiki I can see a few that may be relevant: ACTIVITY_LEVEL, DUTIFULNESS, and CAUTIOUSNESS all seem likely candidates, and a few of the others (ORDERLINESS, ACHIEVEMENT_STRIVING) may also have some bearing.

Loading up Experiment 3 (which uses the same dwarves as the previous experiments), Dastot (Unskilled) has low caution (38) and high excellence (68), while his squadmate Etur has a pretty high nervousness (75) which didn't stop him leading more demonstrations (higher Teacher skill at the end of the year). iton has low caution (31) which, coupled with Cog's disorganized (39, stat is 'organized' so low = disorganized) nature may explain why they attained differing skill levels. Domas, the Axe/Teacher who has usually done pretty well, has a high excellence (67) and is unusually willing to help others (75), and doesn't care what anyone thinks about it either (neurosis of 5).

None of the dwarves in the test environment have unusual (ie. reportable via messages or recorded in Dwarf Therapist's view) activity level or dutifulness attributes.

Going back to the very first save to check their attributes (unfortunately I don't have a tool to get at the exact values at the moment :( ) Dastot shows as 'agile', Etur is 'slow to tire, agile, susceptible to disease', Domas is 'slow to tire, susceptible to disease', Nil (who sat out the last experiment) is 'indefatigable and quick to heal', iton is 'quick to heal and weak', and Cog is 'agile and tough'. (Ral has no physical attributes listed in the beginning)

I know agility is often one considered of the most important stats because it affects speed of action, and I always love watching my highly-trained militia zipping around at about twice the speed of your average dwarf - perhaps having two 'agile' dwarves together did some good here?

Dastot also has a 'feel for music' so he will be the militia's bugler once playable instruments are implemented.

Hopefully, when I run a test for Individual Drills, it should highlight if there's any sort of individual bias caused by particular traits, although running a set of identical tests with a fresh set of dwarves should help to show up variances as well. Something to add to the list I guess! :)
Logged

Werdna

  • Bay Watcher
  • Mad Overlord
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #53 on: April 30, 2012, 03:34:25 pm »

I suspect agile = more attacks in sparring = more skill gains.
Logged
ProvingGrounds was merely a setback.

chewie

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #54 on: April 30, 2012, 06:36:19 pm »

Someone needs to make awesome graphs in excel out of all this information.

Also, PTW.
Logged

MrButtchicker

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #55 on: April 30, 2012, 08:44:43 pm »

I tired this, and it works, like well, for a change. It made military easy, I have many, many, many citizens all with good military skills, just in case something happens. Not a single thief made it past my epic militia yet.

Also, might I add, the teacher skill is hard to train up at first, but good teachers advance quickly. Now just to make sure none of my instructors die.

I would suggest that people put demonstrators through sparring with every type of weapon so the demonstrator can learn all the skills, then the process won't be slowed.
« Last Edit: April 30, 2012, 10:18:59 pm by MrButtchicker »
Logged

expwnent

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #56 on: May 01, 2012, 07:51:12 pm »

Fascinating. I'll have to try and replicate the results at some point.
Logged

Hanslanda

  • Bay Watcher
  • Baal's More Evil American Twin
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #57 on: May 01, 2012, 11:13:16 pm »

Hi, I'm new to the forums, but I've been steadily learning DF through the varied resources available, and I have to say, this single thread has contributed greatly to my mastery of the military.  :) Usage of teacher skill in my current fortress has had... Impressive results.  I expected, (as previous experience had taught) that my three dwarf squad would get slaughtered by the squad of goblin macemen hung up in my grinder entrance, but a martial trance and nine dead goblins later... Praise be to Armok and !!Science!!.
Logged
Well, we could put two and two together and write a book: "The Shit that Hans and Max Did: You Won't Believe This Shit."
He's fucking with us.

MrButtchicker

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #58 on: May 02, 2012, 12:46:33 am »

Woo I tried this, and tested this against normal training in squads of two. My results were better with normal sparring, but they never leveled up in armor using, although they used armor. So if you have a cave spider trapped to train armor using, sparring is the way to go.
Logged

Calech

  • Bay Watcher
    • View Profile
Re: Drunken masters - how to use the Teacher skill
« Reply #59 on: May 03, 2012, 10:22:04 am »

I've finally run the 'control group' experiment, with just Individual Combat Drills, and the results aren't exactly surprising.


The summary - Individual Drills give experience at a fairly slow rate (about 300xp per month, obviously with some variance for dwarven behaviour), but they are steady and consistent. It's most interesting to note that the Individual Drills hit Novice level at about the same time as the Unskilled dwarf squad did - this seems to suggest (as noted above) that if you have no relevant skills at all, it takes about the same length of time in a squad as it does alone. Once some skills are present, the rate of gain from squad activities exceeds that of Individual Drills.

It should be noted that during these experiments I haven't taken into account armour or shields, both of which are also trainable skills, but not usually ones that I consider at embark - this may be an oversight as weapon skills train far faster than the auxiliary combat skills (Dodger, Armor User, Shield User, plus the non-weapon skills Biter, Kicker, Striker and Misc. Obj. User - although only the first three tend to be desirable in most situations).
Logged
Pages: 1 2 3 [4] 5 6